EP0498501B1

EP0498501B1 - Encoding/decoding circuit, and digital video system comprising the circuit

Info

Publication number: EP0498501B1
Application number: EP92200268A
Authority: EP
Inventors: Adrianus Johannes Maria Denissen; Bernardus Antonius Maria Zwaans
Original assignee: Koninklijke Philips Electronics NV; Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 1991-02-07
Filing date: 1992-01-31
Publication date: 1997-04-02
Anticipated expiration: 2012-01-31
Also published as: US5469448A; KR100250590B1; ATE151191T1; DE69218648D1; JP3255681B2; DE69218648T2; KR920017378A; NL9100218A; JPH04320114A; EP0498501A1

Abstract

An integrated circuit for encoding digital data in the recording mode and decoding digital data in the reproduction mode using an error correction product code can be used in combination with a standard frame memory when use is made of a high-frequency system clock. An error correction strategy suitable for digital video is implemented in the circuit.

Description

The invention relates to a circuit for the encoding and decoding of digital data with an error correction code.
In a circuit of this kind, digital data words are formed in a stream of digital data, each data word being encoded according to a given error correction code (i.e. provided with so-called parity symbols) after which they are stored as code words on a storage medium (upon recording). Should the code words no longer correspond completely to the original code words, due to disturbances or damaging of any kind after recovery from the storage medium (upon reproduction), the redundant information in the parity symbols enables error correction upon decoding. It is often advantageous to use a combination of two error correction codes. To this end, the digital data symbols (such as bytes) can be arranged, for example, in rectangular blocks, so that data words are formed in the horizontal and the vertical direction. The horizontal data words on the one side and the vertical data words on the other side can then be protected by way of two (possibly different) codes. The codes together are also referred to as a product code. For encoding and decoding it is necessary to store the digital data temporarily block-wise in an external frame memory. In order to enable simple and fast reading and writing in such an external frame memory, accompanied by a low power dissipation and limited use of wiring, it is advantageous when one (standard) memory suffices.
The invention inter alia has for its object to provide an encoding/decoding circuit which requires a small external frame memory. To achieve this, the invention as defined in claim 1 provides an encoding/decoding circuit which comprises a first sub-circuit for encoding data in the recording mode and decoding data in the reproduction mode according to a first error correction code, a second sub-circuit for encoding data in the recording mode and decoding data in the reproduction mode according to a second error correction code, a third sub-circuit for the supply of non-encoded data in the recording mode and the output of decoded data in the reproduction mode, the first sub-circuit being suitable to output encoded data in the recording mode and to supply encoded data in the reproduction mode, each sub-circuit being suitable to receive clock signals from its own system clock having a frequency, the circuit also comprising switching means which are connected to said sub-circuits and which are controlled by a control circuit for cyclically connecting the sub-circuits, via the switching means, to a terminal for data transfer from and to an external frame memory at a frequency which is at least equal to the sum of the frequencies of the system clocks of the sub-circuits. As a result of the use of fast switching means, for example multiplexers which are controlled at a frequency which is at least equal to the sum of the frequencies of the system clocks of the sub-circuits, it is achieved that all three sub-circuits can communiate as if it were parallel with a single external memory. The control circuit cyclically connects each time one sub-circuit to the external memory, but at such a high frequency that all sub-circuits can access to the memory for a sufficiently long period of time. A cycle is then determined by the frequencies of the system clocks. Therefore, the sub-circuits are not necessarily alternately connected to the memory.

Fig. 1 shows a digital video system suitable for use of a circuit in accordance with the invention;
Fig. 2 shows the encoding/decoding circuit in accordance with the invention;
Fig. 3 illustrates the use of the external frame memory.

Fig. 1 shows a digital video system suitable for use of a circuit in accordance with the invention. A digital image pick up unit 1.1, for example a CCD camera generating 25 frames (50 fields) per second, each with approximately 420.000 pixels, applies digital video data to a video processor 1.5. In the video processor each time two fields applied are combined so as to form one frame which is stored in a memory 1.7, for example a 5-Mbit DRAM. Per pixel the video data contains 8 bits of luminance information (enabling representation of 256 grey values) and chrominance information. Generally, the raster of the chrominance information is coarser than that the luminance information, for example one chrominance value, determined by two 8-bit values, per block of 2x2 pixels. In the video processor 1.5 there are also formed blocks of 8x8 pixels, so called DCT blocks, which can be distinguished as luminance DCT blocks and chrominance DCT blocks. Four luminance DCT blocks in conjunction with their two corresponding chrominance DCT blocks form a so-called DCT unit. Each time five DCT units, obtained in the video processor 1.5 by shuffle, constitute a so-called segment. Shuffle (mixing) has an averageing effect, which is advantageous for the data reduction to be subsequently executed. Each DCT block is transformed in a known Discrete Cosine Transformation (and inverse DCT) circuit 1.9. Subsequently, per segment (i.e. 30 DCT blocks) a data reduction is performed in a known variable-length encoding/decoding circuit 1.10. Therein, 30*64*8=15,360 bits are reduced to 3072 bits, using known techniques such as quantisation and variable-length encoding. Quantisation can be performed, if desired, in various ways in parallel, each time the most suitable way being chosen.
Via an analog-to-digital converter 1.16 one or more microphones 1.3 apply digital audio data to the audio data processor 1.6 which is connected to a memory 1.8, for example a 256-Kbit SRAM. Generally speaking, the amount of audio data is much smaller than the amount of video data, so that the audio data need not be reduced. The audio data of the processor 1.6 (in which the blocks are also formed) and the reduced video data of the circuit 1.10 are applied to an error correction encoding/decoding circuit 1.11 in which the data is provided with a known error correction code, for example a Reed-Solomon product code. A code of this kind is described in United States Patent Specification 4,802,173. A number of bytes (or other data symbols) to be protected by the error correction code is arranged in a rectangular array, after which so-called parity symbols are added to each horizontal row and each vertical column (i.e. to each horizontal and vertical data word) in accordance with the rules of the relevant code. The data words with the added parity symbols are referred to as code words. These parity symbols enable, utilising the redundant information they represent, the correction of errors in the array of bytes which occur during the transport of the data or due to the damaging of the storage medium. In this respect reference is also made to Richard E. Blahut: "Theory and practice of error control codes", 1983, Addison-Wesley Publ. Comp. Inc., or N. Glover & T. Dudley: "Practical error correction design for engineers", 1982, Data System Technology Corp., Broomfield Colorado.
The number of luminance DCT blocks per frame amounts to 720/8*576/8 = 90*72 = 6480. Per frame there are 360/8*288/8*2 = 3240 chrominance DCT blocks. Thus, there are 1620 DCT units per frame, or 324 segments per frame. In the case of a field frequency of 50 Hz, a frame is divided into 12 so-called tracks (in the case of a field frequency of 60 Hz, a frame is divided into 10 tracks). Such a track contains, in addition to the data symbols and the parity symbols, inter alia identification and synchronisation information. The data symbols and the parity symbols of a track are referred to hereinafter as a RS video block. Per video block, therefore, there are 27 segments. Each segment of 3072 bits (= 384 bytes) forms 3 lines of 128 bytes in an RS video block. This includes 1 byte of auxiliary data AUX per line, for example, data concerning the elapsed time or the frame number. An RS video block thus comprises 81 lines of 128 bytes which form 81 horizontal data words and 128 vertical data words. Subsequently, in the circuit 1.11 the parity symbols associated with the constituent data symbols are determined for each RS video block in accordance with a Reed-Solomom product code and are added. For example, first an outer code can be applied to the vertical data words and subsequently an inner code to the horizontal data words. The standard notation for such a Reed Solomon code is RS(k+p,k,p+1), where k is the number of data symbols of the code to be protected and p is the number of parity symbols. For the outer code use can be made of RS(88,81,8), code over GF(256) and derived from the primitive polynomial x⁸+x⁴+x³+x²+1. For the inner code use can be made of: RS(136,128,9), code over GF(256) and derived from the primitive polynomial x⁸+x⁴+x³+x²+1.
To this end, the circuit 1.11 is connected to a memory 1.12, for example a 1 Mbit SRAM. The data (code words) thus encoded in the recording mode is applied to a known modulator/demodulator circuit 1.13 which modulates the data for supply to two read/write heads 1.14 and 1.15 for recording the data on, for example a magnetic tape.
For the reconstruction of the video and audio signals, the two read/write heads 1.14 and 1.15 read the code words stored on the magnetic tape and apply these words to a modulator/demodulator circuit 1.13 after which the demodulated data is corrected in the circuit 1.11 by means of the parity symbols added on the basis of the Reed-Solomon product code. First the inner code (applied to the horizontal video and audio code words) is then decoded, after which the corrected data is separated into audio data, applied to the audio processor 1.6, and video data which is subsequently decoded according to the outer code, after which it is applied to the variable length encoding/decoding circuit 1.10. The audio processor 1.6 applies an audio data, via a digital-to-analog converter 1.17, to the audio output 1.4. The variable length decoding/encoding circuit 1.10 then decodes the variable length code, thus supplementing the amount of data bits obtained to 15,360. Subsequently, this data is applied to the DCT circuit 1.9 in which the inverse Discrete Cosine Transformation is performed. The data thus produced is applied to the video processor 1.5 which renders the data suitable for and applies it to the video output 1.2.
Fig. 2 is a detailed representation of the error correction encoding/decoding circuit 1.11 in accordance with the invention. It comprises a first subcircuit 2.1 with a first encoding circuit 2.2 for encoding data, in the recording mode, according to a first error correction code, a first decoding circuit 2.3 for decoding, in the reproduction mode, data encoded according to the first error correction code, and a first address counter 2.4; a second sub-circuit 2.5 which comprises a second encoding circuit 2.6 for encoding, in the recording mode, data according to a second error correction code, a second decoding circuit 2.7 for decoding, in the reproduction mode, data encoded according to the second error correction code, and a second address counter 2.8; a third sub-circuit 2.9 which comprises an interface circuit 2.10 for the supply of non-encoded data in the recording mode and the output of decoded data in the reproduction mode, and a third address counter 2.11. The structure of the circuit 1.11 is such that the external memory 1.12 can be as small as possible and the necessary access to this memory is minimised. To achieve this, the circuit furthermore comprises switching means 2.12 and 2.13 which connect the sub-circuits to the external memory 1.12 under the control of a control circuit 2.15. The switching means 2.12 connect the address counters 2.4, 2.8 and 2.11 to the address input and output AD of the memory 1.12 and the switching means 2.13 connect the encoding circuit 2.2, the decoding circuit 2.3, the encoding circuit 2.6, the decoding circuit 2.7 and the interface circuit 2.10 to the data input and output I/O of the memory 1.12. Each sub-circuit is controlled by its own system clock. The system clocks are derived, for example from an external clock signal of 13.5 MHz, each system clock amounting to 4.5 MHz and being phase-shifted with respect to one another, for example through 0, 120 and 240 degrees. Thus, together they determine the frequency of 13.5 MHz used to control the switching means and the external memory by the control circuit 2.15 which receives the external clock signal, generates the system clock and the data stream and also ensures correct addressing in the circuit 1.11 and in the external memory 1.12 by means of the address counters 2.4, 2.8 and 2.11.
In the recording mode the data stream is controlled by the control circuit 2.15 as follows. The variable-length encoding/decoding circuit 1.10 applies non-encoded (as regards the error correction code to be added) data to the interface circuit 2.10 , which data is row-wise written into the external SRAM 1.12, under the control of the 4.5 MHz system clock of the third sub-circuit, via the switching means. Under the control of the control circuit 2.15, the address counter 2.11 then acts as a pointer in the external memory. Any auxiliary data AUX can also be applied to the external memory via the interface circuit 2.10. As described above, the data symbols are arranged in 81 rows and 128 columns of bytes per RS video block, so that 81 horizontal data words of 128 bytes and 128 vertical data words of 81 bytes are formed per RS video block 81. Under the control of the 4.5 MHz system clock of the second sub-circuit, a vertical data word is read from the external SRAM, after which this data word is encoded in the second encoding circuit 2.6 according to the second (outer) error correction code (i.e. the associated parity symbols are determined), after which the calculated (vertical) parity symbols associated with the second error correction code are written into the external SRAM, so that the entire code word associated with the relevant data word is stored in the SRAM. Under the control of the control circuit 2.15, the address counter 2.8 then acts as a further pointer in the external memory. Under the control of the 4.5 MHz system clock of the first sub-circuit, a horizontal data word is read from the external SRAM, after which this data word is encoded according to the first (inner) error correction code in the first encoding circuit 2.2, after which the data word, together with the calculated (horizontal) parity symbols associated with the first error corection code (i.e. the horizontal code word), is not stored but applied directly to the modulator/demodulator circuit 1.13. Under the control of the control circuit 2.15 the address counter 2.4 then acts as a further pointer in the external memory. The previously determined vertical parity symbols of the second (outer) error correction code also form horizontal data words which are thus processed in the first encoding circuit 2.2. In alternation with this processing of video data in the circuit 2.2, audio data is also processed (for example in a multiplex fashion): the audio processor 1.6 applies audio data to the circuit 2.2 which provides this data with parity symbols associated with the first error correction code, after which the data is applied, together with the associated, calculated parity symbols, to the modulator/demodulator circuit 1.13.
In the reproduction mode the data stream is as follows. The modulator/demodulator circuit 1.13 applies data to the first decoding circuit 2.3 which determines, on the basis of the identification information generated in and despatched by the circuit 1.13, whether the encoded horizontal data word (i.e. horizontal code word) concerns audio data or video data. Audio data is decoded according to the first (inner) error correction code and is applied to the audio processor 1.6; video data is decoded according to the first (inner) error correction code and, if correctable, it is written into the external memory 1.12 under the control of the 4.5 MHz system clock of the first sub-circuit. Under the control of the control circuit 2.15, the address counter 2.4 then acts as a pointer in the external memory. The addresses at which the data is stored in the memory are contained in the identification information generated in the circuit 1.13. Under the control of the 4.5 MHz system clock of the second sub-circuit, the second decoding circuit 2.7 reads vertical encoded data words (i.e. vertical video code words) from the external SRAM, decodes these words according to the second error correction code, and writes only those data symbols of each relevant data word into the external memory 1.12 that have been corrected by decoding. Under the control of the control circuit 2.15, the address counter 2.8 then acts as a further pointer in the external memory. Under the control of the 4.5 MHz system clock of the third sub-circuit, the decoded data is read from the external memory 1.12 and applied to the variable-length encoding/decoding circuit 1.10 or is output as auxiliary data AUX. The address counter 2.11 then acts as a further pointer in the external memory under the control of the control circuit 2.15.
A storage capacity of 128*81*12*8=995328 bits (# bytes per line times # lines per RS video block times # RS video blocks per frame # bits per byte) is required for the storage of a complete frame in the external memory. A 1-Mbit SRAM can be used as a standard memory; this is approximately 1049 Kbits. The remaining 53 Kbit is used for the storage of the further parity symbols of 6 RS video blocks of the relevant frame, see Fig. 3. The address counters or pointers A1, A2 and A3 (associated with the respective sub-circuits) cyclically and simultaneously scan the external memory, the control circuit 2.15 ensuring that the pointers do not overtake one another and hence can proceed in the rhythm of their system clock. The simultaneous cyclic scanning of the external memory under the control of the control circuit 2.15 takes place in the reverse order during the recording mode and the reproduction mode. Moreover, the vertical parity symbols of 6 RS video blocks are cyclically stored in the appropriate storage spaces VP1/7 to VP6/12. The storage space VP1/7, for example contains the parity symbols of the RS video block 1 or of the RS video block 7, depending on the phase in the cycle of the address counters. The control circuit 2.15 ensures that in the recording mode the address counter A2 erases the relevant vertical parity symbols only after they have been read by the address counter A1, and that in the reproduction mode the address counter A1 erases the relevant vertical parity symbols only after they have been read by the address counter A2. The specification of the 1 Mbit SRAM to be used is: 128 Kbit * 8, frequency 13.5 MHz, bidirectional 8-bit data bus.
As has already been described, the Reed-Solomon product code used employs an outer code which operates on the vertical data words and an inner code which operates on the horizontal data words. The inner code has 8 parity symbols and the outer code has 7 parity symbols. The error correction strategy is as follows. The inner code serves to correct single bit errors (incidental errors) and to detect burst errors (string of bit errors). The outer code serves to correct horizontal detected errors (erasures) and to correct any miscorrections by the horizontal decoder. To this end, 6 of the 8 parity symbols of the inner code can be used to correct at the most 3 errors. The remaining 2 parity symbols then serve as an additional detection margin. Thus, the results of the correction calculation (6 equations with 6 unknowns) are inserted in the remaining seventh and eighth equation for checking purposes, so that additional certainty is obtained as regards the correctness of the calculated corrections. The probability of a miscorrection then amounts to approximately 3*10^-7 for the horizontal decoder (see, for example R.J. McEliece & L.Swanson: "On the decoder error probability for Reed-Solomon codes", IEEE Transactions on Inform. Th., IT-32,5, 1986, pp. 701-703). When more than 3 errors are detected in a horizontal data word, evidently a burst error is concerned. In that case the entire data word is not written into the SRAM, but is made an erasure in its entirety by means of a correction bit as will be described hereinafter. The corresponding preceding data word from the previous frame then remains in the memory. Alternatively, 4 of the 8 parity symbols of the inner code can be used for the correction of at the most 2 errors.In that case the remaining 4 parity symbols are used as an additional detection margin. The 7 parity symbols of the outer code can all be used for the correction of at the most 7 detected erasures. An other possibility consists in that 2 parity symbols are used for the correction of 1 error and the remaining 5 parity symbols are used for the correction of detected erasures. Any miscorrection occurring can thus also be corrected. 1 parity symbol can also be used as an additional detection margin.
The inner code and the outer code thus cooperate: the inner code corrects small random errors and detects burst errors so that the outer code already knows about the location of the erasures corrected. In order to enable such cooperation between the inner and the outer decoder, the circuit 1.11 comprises an embedded SRAM 2.14 (see Fig. 2), connected to the address counters 2.4 and 2.8, the encoding circuits 2.2 and 2.6, and the decoding circuits 2.3 and 2.7 of the first and the second sub-circuit. In the reproduction mode, after the decoding of the video data according to the first (inner) error correction code in the circuit 2.3 and the writing of only the correctable (corrected or errorless) horizontal data words into the memory 1.12, a so-called correction bit is stored in the memory 2.14 for each horizontal data word. This correction bit indicates whether or not the relevant data word was correctable by means of the inner code (corrected or errorless and hence now correct). Any miscorrections, for which the probability of occurrence is very small, are ignored in this respect. When the data word is correct, the correction bit has the logic value 1 and otherwise the logic value 0. Each correction bit having the value 0 is thus associated with a horizontal data word consisting of erasures. When the second decoding circuit 2.7 reads the vertical code words from the external memory in the reproduction mode, the correction bits associated with the horizontal data words are also read from the memory 2.14. The outer decoding circuit thus knows where the erasures to be corrected are located. In the memory 2.14 it is also memorised for the vertical data words whether they are correct. When the data word is correct the correction bit has the logic value 1 and otherwise the logic value 0. The control circuit can thus provide a reliability indication for each byte: if the horizontal data word or the vertical data word containing the relevant byte is correct (i.e. if at least one of the two correction bits of relevance to the byte has the logic value 1), the byte is reliable. This reliability information can be applied, via the circuit 2.10, to the variable length encoding/decoding circuit 1.10. Moreover, when a given threshold value concerning the number of unreliable bytes is exceeded, an indication signal can be generated to indicate that the quality of the tape or the heads is insufficient.
In the reproduction mode, time can be saved as follows during the decoding of the vertical code words by the second (outer) decoding circuit 2.7. When the Reed-Solomon decoder circuit used generates the difference value between the correct value and the incorrect value during the correction of an erasure (i.e. the calculation of the correct value of an incorrect byte whose location is known), it is advantageous to enter the value "O" into the decoding circuit. This is because the circuit then generates directly the correct value which need merely be written into the external memory.
It is to be noted that when use is made of an external SRAM 1.12 which cannot be operated at a clock frequency as high as 13.5 MHz, it is possible to use a 16-bit data bus structure instead of an 8-bit data bus structure, so that the frequency can be halved. In that case, for example, two standard 1-Mbit memories can be used in parallel using the same addressing, the 8 bits presented by the sub-circuits then being buffered as 16 bits which are each time split into 8-bit portions by the switching means 2.13, said portions being applied to the respective data inputs and outputs I/O of the two memories at half the frequency. Conversely, the 8-bit portions supplied at half the frequency by the two memories are then combined to 16 bits by the switching means, each time 8 bits thereof being applied to the sub-circuits at the original frequency.

Claims

A circuit being arranged for bidirectional attachment between a user channel (1.10) and a system channel (1.13), for encoding user data symbols by means of a product code to error protected system data symbols, as well as for decoding such error protected system data symbols to user data symbols while executing an error correction operation,
said circuit having a first subcircuit (2.1) attached to said system channel (1.13) for decoding first layer words of said product code for presentation to a local read-write memory (1.12) whilst producing intermediate error protectivity data, a second subcircuit (2.5) for decoding second layer code words of said product code accessed in said read write-memory (1.12) and on the basis of said intermediate error protectivity data, and a third subcircuit (2.9) for accessing decoded user symbols for presentation onto said user channel,

said third sub-circuit (2.9) being arranged for receiving user symbols from said user channel for presentation to said local read-write memory (1.12), said second sub-circuit (2.5) being arranged for accessing said user symbols in said local read-write memory (1.12) for therefrom constituting second layer code words by means of added redundancy symbols for presentation to said read-write memory (1.12), said first sub-circuit (2.1) being arranged for accessing said second layer code words in said local read-write memory (1.12) for therefrom constituting first layer code words by means of further added redundancy symbols for presentation to said system channel (1.13),

said circuit furthermore comprising control means (2.15) for controlling mutually exclusive access cycles to said local read-write memory (1.12) among said first, second and third sub-circuits at a uniform control frequency that is at least equal to the sum of the maximum access frequencies of said first, second and third sub-circuits.
A circuit as claimed in Claim 1, wherein said maximum access frequencies are equal among said first, second, and third sub-circuits.
A circuit as claimed in Claim 2, wherein said access cycles are cyclically assigned to said first, second and third subcircuits in a first order at encoding, and in a second order opposed thereto at decoding.
A circuit as claimed in Claims 1, 2 or 3, wherein said first sub-circuit (2.1) is arranged for presenting said intermediate error protectivity data to an embedded memory (2.14) for subsequent accessing by said second sub-circuit (2.5).